School Fee Abolition and its Effects on Education Indicators Ray Langsten Social Research Center ...

25 Pages·2014·0.3 MB·English

School Fee Abolition and its Effects on Education Indicators

Ray Langsten
Social Research Center
American University in Cairo
P.O. Box 74
New Cairo, EGYPT 11835
email: [email protected]

March 2014

Abstract
The international community has declared that fee abolition is an important part of
achieving Universal Primary Education (UPE) and that fee abolition reduces differentials in
enrollment and attainment by gender and wealth. I use DHS data to examine the impact of fee
abolition on the Net Enrollment Rate (NER) and on the Primary School Completion Rate
(PSCR). The effect of fee abolition on the NER is much greater than its effect on the PSCR.
Moreover, although after fee abolition gender and wealth differentials in the NER are reduced,
in the PSCR these differentials often remain about the same, or may actually increase. I show
that these conflicting results are found across a range of countries and explain the reasons for
this differential effect.
School Fee Abolition and its Effects on Education Indicators

Introduction / Literature Review

The Education Millennium Development Goal (MDG-goal 2) of Universal Primary
Education (UPE), a reformulation of the second Dakar Education for All (EFA) goal
(UNESCO 2000), states that “by 2015, children everywhere, boys and girls alike, will be able
to complete a full course of primary schooling” (UN Statistics Division 2008; see also
UNESCO 2007: 41). The gender equality MDG (goal 3) mandates elimination of “gender
disparity in primary and secondary education, preferably by 2005” (UN Statistics Division
2008). The School Fee Abolition Initiative (SFAI) has been advanced as a key factor in
spurring progress toward UPE and increasing educational equity (World Bank 2009).
The 2008 EFA Global Monitoring Report (GMR)1 uses the Net Enrollment Rate
(NER) as the critical indicator in “a systematic assessment of progress towards EFA since
Dakar” (UNESCO 2007: 32). The GMR states that “A country’s distance from UPE
appears most clearly in terms of the net enrolment ratio (NER), the share of children of
official primary school age who are actually enrolled in primary schools” (UNESCO 2007:
42 – emphasis added). Over time, the NER, together with the proportion of pupils starting
grade 1 who reach the last grade of primary (the primary school “survival rate”), have come
to be the main indicators of progress toward the education MDG (UN Statistics Division
2008). The gender parity index (GPI) is officially measured by the ratio of gender-specific
Gross Enrollment Ratios (GER) (UNESCO 2007:200), though some researchers focus on

1 The World Education Forum gave UNESCO the responsibility to “co-ordinate the global efforts to achieve”
and “monitor progress toward” EFA (UNESCO n.d.:6). The United Nations and the World Bank also issue
annual reports on the MDGs in general, including basic information about the education MDG. EFA GMRs
can be found here:
http://www.unesco.org/new/en/education/themes/leading-the-international-agenda/efareport/
The most recent UN MDG monitoring report can be found here:
http://unstats.un.org/unsd/mdg/Resources/Static/Products/Progress2012/English2012.pdf
The most recent World Bank MDG monitoring report is here:
http://www-wds.worldbank.org/external/default/WDSContentServer/WDSP/IB/2011/04/25/000386194_201104
25025358/Rendered/PDF/613110PUB0impr187000109780821387009.pdf

1
NERs to assess gender equity (al-Samarrai and Zaman 2007; Deininger 2003).. A GPI
between 0.97 and 1.03 is considered parity.
Early in the new millennium the international community suggested an alternative
indicator for measuring progress toward UPE – the Grade Four Completion Rate (UNESCO
2002b). Given the focus of UPE on primary school completion, an early report urged that
“Primary school completion rates ... should be the criterion for evaluating progress toward the
goal of EFA.” (UNESCO 2001b:42; see also Bruns, Mingat and Rakotomalala 2003;
Carr-Hill 2009; Kane 2004; UNESCO 2003; 2007; 2010). In this paper, in addition to the
NER, I will use a primary school completion rate (PSCR) closely related to the grade-4 and
grade-5 completion rates used in earlier work (Filmer 2005; Lloyd and Blanc 1996; Lloyd,
Kaufman and Hewett 2000). The PSCR is the outcome measure of the proximate
determinants of educational attainment framework (Langsten forthcoming). The proximate
determinants are: ever-enrollment, retention, and timely progress. They are defined so as to
completely determine the PSCR (Langsten forthcoming).
Since 1990, several sub-Sahara African (SSA) countries, along with a few other
countries, have eliminated primary school fees (UNESCO 2007:113 Map 3.2; see also Horn,
Wright and Prouty 2009; Tomasevski 2003). Studies examining the impact of fee abolition
in SSA countries report not just increased enrollment, but also greater socioeconomic and
gender equity following elimination of fees (al-Samarrai and Zaman 2007; Deininger 2003;
World Bank 2009). As long ago as 2007 some observers argued that several African
countries were “on track to reach the MDG of universal completion” by 2015 (Filmer
2007:166). More recently there has been a growing trend in the development community to
declare that “By 2015, the universal primary education … MDG will be met in nearly all
countries” (Beatty and Pritchett 2012:i; also World Bank 2012:12).

2
Beyond this impact on overall educational achievement, fee elimination has been
described as a “pro-poor strategy” (Nishimura, Yamano and Sasaoka 2008; World Bank
2009a). Others, however, maintain that, in the developing world particularly, schooling
underpins the advantages of society’s elites who seek to perpetuate their advantages as
education expands (Carnoy 1974) whether in response to elimination of fees or other policy
changes. Indeed, studies in developed countries, considering other levels of education, have
consistently found that elimination of fees does not increase educational equity (Halsey, Heath
and Ridge 1980; Raftery and Hout 1993).
In this paper I will examine wealth and gender equity in primary education in the
context of fee abolition. I will show that the size of between group differentials, and in some
cases even the direction of change, can differ depending the educational outcome measure used.
I will argue that these inconsistent results occur largely because of the way that the NER, the
GER, and the PSCR respond to timely progress of children through school, one of the
proximate determinants of education attainment.

Data and Methods
Although the NER is usually computed from service statistics, there are survey based
equivalents2. The PSCR is a survey based indicator. Often, even in the EFA Global
Monitoring Reports (GMRs), analyses by gender and especially wealth are carried out using
survey data—usually data from Demographic and Health Surveys (DHS) or Multiple Indicator
Cluster Surveys. In this work I will use data from two SSA countries – Malawi and Uganda –
that eliminated school fees during the 1990s, as well as from Cambodia which eliminated fees

2 The survey based equivalent is technically called the “net attendance ratio” (NAR). Throughout this paper, I
will use the abbreviation NER when I am referring to GMR-reported, administrative-statistic-based figures, and
NAR when discussing survey-based results.

3
in 2000. In Malawi and Uganda there was one DHS before the elimination of fees. In
Cambodia, the first DHS took place in the same year that fees were eliminated – 2000. All of
these countries have at least three surveys with the most recent survey taking place around 2010
and being 10 or more years after fee abolition.
DHS provide all the data necessary to compute the outcome indicators (NAR, GAR and
PSCR) and the survey-based proximate determinants. They also provide two key indirect
determinants of enrollment and attainment: gender and wealth.
In assessing changes in equity we will use very simple numbers. When considering
wealth, we will look at the change in the absolute difference between wealth groups. For
gender we will use the GPI, that is the male to female ratio of the indicator in question.

Results
Overall Educational Attainment
By way of background I first consider change over time in educational indicators –
the GMR reported NERs, NARs based on DHS, and the PSCR. For Malawi and Uganda I
also include NARs reported by others who have assessed the impact of fee abolition in those
countries using other data sets (Malawi – al-Samarrai and Zaman 2007; Uganda – Deininger
2003). I report two PSCRs: 1) a standard PSCR for an age group slightly older than the
normal age of completing primary (see Langsten Forthcoming); and 2) a “cohort” or
completed PSCR. This latter measure estimates the PSCR for the same cohort used in one
DHS, aged to the following DHS. For example: consider the cohort PSCR for the 2010
Malawi DHS (41%). This is based on children aged 22-23 in 2010. That is the same
cohort that was ages 16-17 in 2004, six years earlier. Similarly the cohort PSCR in 2004
(41%) is for children aged 20-21; that is children who were 16-17 years old, four years earlier

4
in the 2000 DHS. The cohort PSCR is useful because in most SSA countries, large numbers
of children are greatly delayed in their progress through school and in completing primary
education. Low standard PSCRs reflect this delay. By the age of the cohort PSCR almost
all children in the age group have either completed primary school or dropped out. Thus the
cohort PSCR can be thought of as the “completed” PSCR for that age group.
These educational indicators are shown in Tables 1-3 for Malawi, Uganda and
Cambodia, respectively. In general all the NERs and NARs are low during the early 1990s.
In the late 1990s, and especially following fee abolition all the NERs & NARs increase –
often substantially. However, post-2000, after fee abolition, the DHS NAR is always lower
than the GMR-reported NER by at least 10 percentage points. (This tendency is confirmed
by an analysis of 21 sub-Saharan African nations. Overwhelmingly, the post-2000, and
where applicable the post-fee-abolition, DHS NARs are lower than, and often substantially
lower than, the GMR-reported NERs. [Data not shown.])
The PSCRs are lower, often very much lower, than even the DHS-NARs. This is
true even of the cohort PSCRs. In Malawi, post-2000, the NAR-PSCR differences are on
the order of 40-50 percentage points. In Uganda and Cambodia these differences are
smaller, but in the range of 15 to 30 percentage points for the NAR – cohort PSCR divide.
The reason for these large differentials can be found in the different way in which the
NAR and PSCR respond to the proximate determinants of educational attainment
(ever-enrollment, retention, timely-progress). These effects have been explained in detail
elsewhere (Langsten Forthcoming). Briefly, the main cause of the differential is the low
level of timely progress that characterizes these three, and many other, developing countries.
Delayed progress through the years of primary education is not captured in the NER/NAR.
Extreme delay, such as is seen in Malawi, results in high NARs but very low PSCRs. Many

5
children remain in school in the age group used for the standard PSCR without having yet
completed primary education. In subsequent years these delayed children have relatively
high dropout rates. Though many delayed children do complete primary, even the cohort
PSCR is low, as noted above. As a general rule of thumb: the more that pre-completion
dropout is delayed , the greater will be the difference between the NER/NAR and the PSCR.

Wealth
Fee abolition, especially in the context of educational expansion has been seen as
“pro-poor” (Nishimura, Yamano and Sasaoka 2008; World Bank 2009a). In Malawi and
Uganda research has examined absolute differences in NARs by wealth quintile before and
after fee abolition (al-Samarrai and Zaman 2007; Deininger 2003). Following fee abolition,
levels of attendance among the less well-off and the poor converge on the attendance level of
children from the wealthiest families (Tables 4 & 5 – Columns 5 & 6). These results are
replicated using DHS data. As an example, in Malawi in 1992, the absolute difference
between the NAR of the rich (R – 82%) and that of the middle wealth group (M – 56%) is 26
percentage points, while this difference between the rich and the poor (R-P) is 42 percentage
points. In Malawi, fees were eliminated in 1994 (Al-Samarrai and Zaman 2007; Inoue and
Oketch 2008). In the three DHS in 2000 and later, these R-M and R-P differences are much
smaller.
Because the standard PSCR in Malawi is based on children 16-17 years of age, the
full effects of fee elimination may not be seen until 2004. However, it is precisely in this
year that the R-M and R-P differences are the largest (42 and 47 percentage points
respectively). While there is some convergence between 2004 and 2010, still in 2010 the
wealth differentials are substantially larger than they were in 1992 before fee abolition.

6
In addition, the wealth differentials tend to grow as the cohort ages. That is, among
children 16-17 years of age from rich families in 2000, 44% had completed primary
education. By 2004, among this rich cohort, now aged 20-21 years old, 76% had completed
primary, an increase of 32 percentage points in primary completion over the 4 years between
surveys. By contrast, children from the middle wealth group increased their level of
primary completion by just 4 percentage points while children from poor families increased
their completion by 17 percentage points. When considering the cohort aged 16-17 years in
2004 and comparing their primary completion with that same cohort aged 22-23 in 2010, we
see a slight convergence in the R-M difference (42 percentage points declining to 38
percentage points), but the R-P difference continues to grow (47 to 54 percentage points).
Uganda shows similar, if less dramatic results. As in Malawi, whether using the
NARs reported by Deininger (2003) or those based on DHS data, the wealth differentials in
the pre-fee-abolition NARs are relatively large. Following fee abolition in 1996 all of the
wealth differentials shown in columns 5 and 6 of Table 5 are considerably smaller. Also, as
in Malawi, the pattern for the PSCRs is the inverse. Following fee abolition, the wealth
differentials are larger than before fee abolition. And the rich – poor cohort PSCR
differentials are as large as, or larger than the differentials for the standard PSCRs. Only in
the case of the rich – middle comparisons do the differentials shrink as the cohort ages.
In Cambodia (Table 6), fees were eliminated in 2000, the same year as the first DHS.
As in the other countries, the DHS NARs are relatively high in 2000 and substantially lower
in the two subsequent surveys. And, as before, the PSCR differences, at least between
children from rich and poor families, are very large in 2000, and remain, at least as large in
the two later surveys. In Cambodia, the rich – middle differentials do converge. This is
likely the result of ceiling effects, with children of the wealthy having very high completion

7
rates by 2010, and trickle down of educational advantages. For the poor, even for the cohort
PSCR the gap with the rich remains large, despite ceiling effects. Trickle down has not yet
reached the poor.
As with the overall differences between the NARs and PSCRs shown in Tables 1-3,
these differences in the patterns of convergence/divergence in the wealth-specific PSCRs
over time are largely driven by differentials and trends in the rate of timely progress. In all
these countries the rate of timely progress is very low – especially in the early surveys.
Perhaps surprisingly, the rate of ever-enrollment is relatively high – ranging from 78% to
90% even in the first survey in the series, increasing to 96% - 97% in all three countries in the
most recent surveys (2010 – 2011). The level of retention tends to fall between these other
two indicators. Wealth differentials in the proximate determinants tend to follow a similar
pattern – relatively small for ever-enrollment, very large for timely-progress, in the middle
for retention. All the differentials shrink over time, but the rich – poor differentials in
timely progress are particularly resistant to change – thus accounting for the persistence of
the rich – poor gaps in completion. And because children of the poor have lower rates of
retention in the ages following the age of the standard PSCR, the rich poor gap tends to
persist, or even increase, as a cohort ages from one survey to the next.

Gender
Change over time in gender differences in various educational indicators is particularly
interesting. They again illustrate the effects of timely progress. According the GMRs the
gender parity index is the ratio of the gender-specific gross enrollment ratios (UNESCO
2007:200). However, research on gender parity yields contradictory results depending on
the indicator used to measure the GPI (al-Samarrai and Zaman 2007; Bennell, Bulwani and

8
Musikanga 2008). It has long been recognized that the NER is more likely to show a female
advantage in attendance than does the GER (UNESCO 2003:49). The NARs and GARs in
Tables 7-9 are consistent with this view. In every survey, in all three countries, the NAR
shows a lesser female disadvantage or greater female advantage than does the GAR. Also in
all three countries there is a tendency for the NAR to move over time from a male advantage
to a female advantage, or at least to gender equality, in attendance. In the case of the GAR
too, there is a shift over time to a smaller male advantage. But in all three countries, a
strong male GAR advantage remains in the most recent data. The primary reason for this
inconsistency is that males have lower levels of timely progress than females. Thus males
whose progress through school has been delayed, and who are over the normal age of primary
attendance will contribute to the apparent male advantage in the GAR, but will be excluded
from the NAR, allowing female attendance to be assessed on a more equal basis.
The effects of delay are more clearly illustrated by the PSCR and cohort-PSCR. In
Malawi the PSCR shifts from a large male advantage in 1992 to a strong female advantage in
all subsequent years. However, as the cohorts age, the male advantage in 1992 grows, while
the female advantages in the 2000 and 2004 surveys disappear and become strong male
advantages in the respective cohort PSCR. While the size of the male advantage in the
cohort PSCR declines over time, it remains very large in the 2010 survey.
Results for Uganda and Cambodia are similar, except that there is female equality or a
substantial female advantage in the PSCR in all surveys. As in Malawi there is a shift
toward a strong male advantage, or at least male skewed equality, as each cohort ages. As in
Malawi the size of the male advantage in the cohort PSCR declines over time.
These gender effects of fee abolition are particularly interesting, because in the case
of gender it is the favored group (males) who have lower levels of timely progress. It

9